Please share how you are using CharLS

Coordinator
Apr 4, 2009 at 1:16 PM
Edited Apr 4, 2009 at 1:17 PM
I have gotten feedback over e-mail from several people that are testing CharLS.

I am curious what people are using it for (or, considering using it for)
So:
- what kind of image sizes,
- what bitdepth,
- color or monochrome images
- source of the images (digital camera, medical equipment, computer generated, desktop screenshots)
- what is the image's pixel format: planar or color interleaved,
- do you use lossy or lossless mode of JPEG-LS
- what platform: linux, windows, mac, 64 or 32 bit


Any feedback would be greatly apriciated!

Jan
Apr 4, 2009 at 1:30 PM
I'm writing a plug-in for Irfanview, a Windows image viewer.

- about up to 500MB of image size
- 8bpp only, on saving, if this is under 8bpp, up-converted to 8bpp.
- both color and monochrome. monochrome image is converted 256 paletted bitmap.

There is a problem about pixel format.

The BITMAP pixels with lining forrowing

Y=n:
BGRBGRBGR...
Y=n-1:
BGRBGRBGR...
..
Y=0:
BGRBGRBGR...

In the library, it is assumed in this way.

Y=0:
RRR...GGG...BBB
Y=1:
RRR...GGG...BBB
....
Y=n:
RRR...GGG...BBB

So, we must sort all pixels before compressing/decompressing.
This needs very large memories!

However, in the library, data is processed for one line to next line.
So, if charls support for bitmap pixel format, we will not need temporary buffers.
Apr 10, 2009 at 12:06 AM
I support this myself.

Here is a patch.
http://knivez.homelinux.org/~spro/vcpp/supportbmp.patch